Lane-crossing intention prediction method, device, electronic device, and autonomous driving vehicle

Abstract

The present disclosure provides a method, device, electronic device, and autonomous vehicle for predicting lane-crossing intentions, relating to the field of artificial intelligence, and more particularly, autonomous driving. A specific implementation scheme comprises: obtaining perception information corresponding to a target vehicle; generating lane-crossing intention information of the target vehicle based on the perception information and a lane-crossing intention judgment strategy; obtaining a lane-crossing intention prediction result for the target vehicle from a prediction module of a lane-crossing prediction model, wherein the prediction module is a machine learning model for predicting whether the target vehicle has a lane-crossing intention by predicting the trajectory of the target vehicle; processing the lane-crossing intention prediction result based on the lane-crossing intention information to obtain a model output result of the lane-crossing prediction model, wherein the model output result matches the lane-crossing intention information.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to the field of autonomous driving. Background Art

[0002] In the field of autonomous driving, it's necessary to predict the trajectories of vehicles ahead and behind the autonomous vehicle. This prediction is then fed into the control system, which generates control instructions to steer the autonomous vehicle. For example, if the vehicle ahead of the autonomous vehicle slows down or crosses lanes, the autonomous vehicle must predict the vehicle's trajectory and avoid it before completing the crossover. Summary of the Invention

[0003] The present disclosure provides a method, apparatus, electronic device, computer product, computer-readable storage medium, and autonomous driving vehicle for predicting lane-crossing intention.

[0004] According to one aspect of the present disclosure, a method for predicting lane-crossing intention is provided, including: obtaining perception information corresponding to a target vehicle; generating lane-crossing intention information of the target vehicle based on the perception information and a lane-crossing intention judgment strategy; obtaining a lane-crossing intention prediction result for the target vehicle by a prediction module of a lane-crossing prediction model, wherein the prediction model is a machine learning model for predicting whether the target vehicle has a lane-crossing intention by predicting the trajectory of the target vehicle; processing the lane-crossing intention prediction result based on the lane-crossing intention information to obtain a model output result of the lane-crossing prediction model, wherein the model output result matches the lane-crossing intention information.

[0005] According to another aspect of the present disclosure, a lane-crossing intention prediction device is provided, including: a first acquisition module for acquiring perception information corresponding to a target vehicle; a generation module for generating the lane-crossing intention information of the target vehicle based on the perception information and a lane-crossing intention judgment strategy; a second acquisition module for acquiring a lane-crossing intention prediction result of a prediction module of a lane-crossing prediction model for the target vehicle, wherein the prediction model is a machine learning model for predicting whether the target vehicle has a lane-crossing intention by predicting the trajectory of the target vehicle; and a processing module for processing the lane-crossing intention prediction result according to the lane-crossing intention information to obtain a model output result of the lane-crossing prediction model, wherein the model output result matches the lane-crossing intention information.

[0006] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute any one of the above-mentioned cross-lane intention prediction methods.

[0007] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the cross-lane intention prediction method according to any one of the above items.

[0008] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, which implements any of the above-mentioned methods for predicting cross-lane intention when executed by a processor.

[0009] According to another aspect of the present disclosure, an autonomous driving vehicle is provided, wherein the autonomous driving vehicle includes the lane-crossing intention prediction device described in any one of the above items.

[0010] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0012] Figure 1 is a flowchart of a method for predicting lane-crossing intentions according to an embodiment of the present disclosure;

[0013] Figure 2 is a schematic diagram of a lane-crossing intention judgment strategy provided according to an optional embodiment of the present disclosure;

[0014] Figure 3 is a schematic diagram of perception information provided according to an optional embodiment of the present disclosure;

[0015] Figure 4 is a structural block diagram of a device for predicting lane-crossing intention according to an embodiment of the present disclosure;

[0016] Figure 5 A schematic block diagram of an example electronic device 500 for implementing an embodiment of the present disclosure is shown;

[0017] Figure 6 1 is a structural block diagram of an autonomous driving vehicle provided according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0018] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0019] In the field of autonomous driving, reasoning about the behavioral intentions of vehicles ahead and behind has always been a crucial aspect of autonomous driving applications. Related technologies typically use machine learning models to predict the trajectories of vehicles ahead and behind. Based on these predicted trajectories, the future behavior of these vehicles can be predicted. However, these machine learning models are trained using sample data, and their interpretability and reliability are often difficult to verify. Predicting the lane-crossing intentions of large vehicles ahead and behind the autonomous vehicle has been a particular challenge. The prediction accuracy and recall rates of machine learning models have consistently struggled to meet requirements, potentially leading to collisions with large vehicles due to incorrect behavior predictions, impacting road safety.

[0020] In order to solve the above problems, this paper proposes a method for predicting lane-crossing intentions that can be applied to autonomous vehicles. Figure 1 is a flow chart of a method for predicting lane-crossing intention according to an embodiment of the present disclosure, such as Figure 1 As shown, the method includes the following steps:

[0021] Step S102: Acquire the perception information corresponding to the target vehicle.

[0022] Optionally, the target vehicle may be a preceding or following vehicle of the current vehicle to which this method is applied, particularly a large vehicle. Perception information corresponding to the target vehicle may be obtained from the current vehicle's perception system. For example, the perception system may include structures such as visual sensors, radar, and a map perception module for obtaining perception information related to the target vehicle. Therefore, the perception information may include visual images, radar images, electronic maps, and the like related to the target vehicle.

[0023] Step S104 : generating lane-crossing intention information of the target vehicle based on the perception information and the lane-crossing intention judgment strategy.

[0024] In this step, the lane-crossing intention judgment strategy can be a strategy pre-installed in the computing system of the current vehicle, which is used to parse the perception information related to the target vehicle according to a strategy with strong interpretability to determine whether the target vehicle wants to cross lanes, that is, whether the target vehicle wants to drive from the lane where the target vehicle is currently located to the lane where the current vehicle is located.

[0025] Step S106: Obtain a prediction result of the lane crossing intention of the target vehicle by a prediction module of the lane crossing prediction model, wherein the prediction module is a machine learning model for predicting whether the target vehicle has the lane crossing intention by predicting the trajectory of the target vehicle.

[0026] It should be noted that the prediction module is a module in the lane crossing prediction model that is used to predict the driving trajectory of the target vehicle and output a lane crossing intention prediction result. Any lane crossing intention prediction result given by the prediction module may include the predicted trajectory of the target vehicle, and the lane crossing intention of the target vehicle corresponding to the lane crossing intention prediction result may be determined based on the predicted trajectory. Optionally, the prediction module may be an architecture that adopts a deep neural network model, or it may be an architecture form of other types of deep learning models or machine learning models. Generally, the lane crossing intention prediction result of the target vehicle predicted by the prediction module may include multiple prediction results, but will not include all prediction results related to the trajectory and lane crossing intention predicted by the machine learning model. For example, the prediction module can predict the target vehicle's trajectory and lane-crossing intention, predicting that the target vehicle intends to cross the lane to the left at a constant speed with a probability of 40%, the target vehicle intends to go straight at a constant speed with a probability of 30%, the target vehicle intends to cross the lane to the right at a constant speed with a probability of 20%, the target vehicle intends to accelerate to the left with a probability of 5%, the target vehicle intends to accelerate to the right with a probability of 2.5%, and the target vehicle intends to accelerate to go straight with a probability of 2.5%. These are all the prediction results related to the trajectory and lane-crossing intention predicted by the prediction module. However, the prediction module does not output all prediction results as lane-crossing intention prediction results. Instead, it outputs the prediction results with higher probabilities as lane-crossing intention prediction results according to certain rules. For example, only the target vehicle intends to cross the lane to the left at a constant speed as the lane-crossing intention prediction result, while pruning the prediction results with lower probabilities. As a result, the prediction results representing the target vehicle's true intention may be pruned, resulting in misjudgment, reducing the recall and accuracy of the prediction results.

[0027] It should be noted that recall refers to the number of correct information extracted by the model divided by the total number of correct information, that is, the number of correct results included in the model output divided by the number of all correct results.

[0028] The above solution adds a processing strategy to the prediction module. This strategy is used to correct the inaccurate and low recall of the lane-crossing intention prediction results output by the prediction module. This solution proposes a lane-crossing prediction logic independent of the machine learning model. Based on this logic, whether the target vehicle intends to cross lanes is determined and lane-crossing intention information is generated. The lane-crossing intention information is then compared with the lane-crossing intention prediction result output by the prediction module to ensure a match. If there is a mismatch, the lane-crossing intention output of the prediction module can be corrected based on the lane-crossing intention information.

[0029] It should be noted that determining whether the model output matches the lane-crossing intention information is to determine whether the prediction result of the prediction module matches the lane-crossing intention information. Optionally, the process of determining whether the prediction result of the prediction module matches the lane-crossing intention information can be illustrated by the following example:

[0030] Each prediction result from the prediction module can include the predicted trajectory of the target vehicle, and the target vehicle's lane-crossing intention corresponding to the prediction result can be determined based on the trajectory. For example, when the predicted trajectory of the target vehicle crosses the lane to the left, the lane-crossing intention corresponding to the prediction result can be determined to be crossing to the left. In this case, if the lane-crossing intention information generated according to the lane-crossing intention judgment strategy indicates that the target vehicle has an intention to cross to the left, then the prediction result from the prediction module can be considered to match the lane-crossing intention information. However, if the lane-crossing intention information indicates that the target vehicle has an intention to cross to the right, or indicates that the target vehicle has no intention to cross, then the prediction result from the prediction module can be considered to not match the lane-crossing intention information.

[0031] Step S108 : Processing the lane crossing intention prediction result according to the lane crossing intention information to obtain a model output result of the lane crossing prediction model, wherein the model output result matches the lane crossing intention information.

[0032] In this step, a match can be made between the lane-crossing intention information and the lane-crossing intention prediction result. When the two do not match, the lane-crossing intention information is prioritized, and the lane-crossing intention prediction result output by the prediction module is corrected. The corrected result is then given as the output of the entire lane-crossing prediction model to the downstream system, such as to the downstream vehicle control system, to help the vehicle control system generate control instructions. It should be noted that for smart cars, the downstream vehicle control system requires the prediction result of a machine learning model that can predict the trajectory of the target vehicle. The prediction result includes the lane-crossing intention and the predicted trajectory, while the lane-crossing intention information only includes information about whether the target vehicle is going to cross lanes, but does not include the predicted trajectory of the target vehicle. Therefore, the downstream vehicle control system cannot apply the lane-crossing intention information. Therefore, it is necessary to correct the lane-crossing intention prediction result based on the lane-crossing intention information, and then give the corrected prediction result to the downstream to complete the intelligent control of the vehicle. Optionally, correcting the cross-lane intention prediction result may include modifying the predicted trajectory of the cross-lane intention prediction result so that the trajectory conforms to the cross-lane intention information, or may include replacing the cross-lane intention prediction result output by the prediction module with a prediction result that conforms to the cross-lane intention information, and then outputting the prediction result that conforms to the cross-lane intention information as a model output result to the downstream system.

[0033] Through the above steps, the purpose of optimizing the prediction results of the target vehicle's lane-crossing intention can be achieved, the technical problem of insufficient recall rate of the machine learning model for reasoning about the vehicle's lane-crossing intention can be solved, and the technical effect of improving the recall rate of the machine learning model's prediction results for the vehicle's lane-crossing intention can be achieved.

[0034] As an optional embodiment, the cross-lane intention prediction results are processed according to the cross-lane intention information to obtain the model output results of the cross-lane prediction model. The following method can be adopted: determine whether there are results in the cross-lane intention prediction results that match the cross-lane intention information; if there are no results in the cross-lane intention prediction results that match the cross-lane intention information, recall the prediction results that match the cross-lane intention information from all the prediction results of the cross-lane prediction model and use them as the model output results of the cross-lane prediction model.

[0035] It should be noted that, among all the prediction results of the cross-lane prediction model, some prediction results are given by the prediction module as cross-lane intention prediction results, and other prediction results may be discarded or pruned because the corresponding probabilities are too low. Therefore, this optional embodiment can recall the prediction results that match the cross-lane intention information from the discarded or pruned prediction results, which is equivalent to retrieving the discarded prediction results and giving them to the downstream system as the model output results of the entire cross-lane prediction model.

[0036] The present disclosure provides the following optional embodiments for ease of understanding: if the prediction module predicts the target vehicle's intention to cross the lane, all the prediction results obtained may include the following items: the probability that the target vehicle intends to cross the lane to the left at a constant speed is 40%, the probability that the target vehicle intends to go straight at a constant speed is 30%, the probability that the target vehicle intends to cross the lane to the right at a constant speed is 20%, the probability that the target vehicle intends to accelerate to cross the lane to the left is 5%, the probability that the target vehicle accelerates to cross the lane to the right is 2.5%, and the probability that the target vehicle accelerates to go straight is 2.5%. The prediction result of the intention to cross the lane output by the prediction module may only include the prediction result of the intention to cross the lane to the left at a constant speed, but the intention to cross the lane information generated based on the intention to cross the lane judgment strategy may support the target vehicle accelerating to cross the lane to the right, that is, the target vehicle accelerating to cross the lane to the right is the true and accurate result. Then, this optional embodiment can recall the prediction result of the target vehicle accelerating to cross the lane to the right, and give it to the downstream system as the model output result of the entire lane crossing prediction model to ensure the recall rate of the model output result of the model.

[0037] As an optional embodiment, the perception information may include at least one of the following: line crossing information, turn signal information, speed angle and vehicle head direction angle; line crossing information is used to characterize whether the target vehicle has crossed the road line, and the road line is the road line close to the current lane included in the lane where the target vehicle is located; turn signal information is used to characterize the turn signal status of the target vehicle; the speed angle is used to characterize the angle between the speed direction of the target vehicle and the road line; the vehicle head direction angle is used to characterize the angle between the vehicle head direction and the road line.

[0038] Based on this optional embodiment, a rich basis for judging whether the target vehicle has the intention to cross the lane can be provided, thereby improving the accuracy of judging the target vehicle's intention to cross the lane.

[0039] As an optional embodiment, the lane-crossing intention judgment strategy may include at least one of a first strategy, a second strategy, and a third strategy, wherein the first strategy is to judge that the target vehicle has the intention to cross lanes when the target vehicle turns on the turn signal in the direction of the current lane, and the speed angle deviates toward the current lane and is greater than a first threshold; the second strategy is to judge that the target vehicle has the intention to cross lanes when the target vehicle passes over the road line, and the speed angle deviates toward the current lane and is greater than a second threshold; the third strategy is to judge that the target vehicle has the intention to cross lanes when the target vehicle meets any one of the first condition, the second condition, and the third condition, wherein the first condition includes: any one of the speed angle and the vehicle head direction angle deviates toward the current lane and is greater than the third threshold; the second condition includes: the target vehicle turns on the turn signal in the direction of the current lane and / or passes over the road line, and any one of the speed angle and the vehicle head direction angle deviates toward the current lane and is greater than a fourth threshold; the third condition includes: the target vehicle turns on the turn signal in the direction of the current lane and passes over the road line, and both the speed angle and the vehicle head direction angle deviate toward the current lane.

[0040] It should be noted that the first threshold can be 5 degrees, the second threshold can be 3 degrees, the third threshold can be 2 degrees, and the fourth threshold can be 1 degree. As the threshold values gradually decrease, the judgment of whether the target vehicle has the intention to cross lanes becomes more relaxed, ensuring that the target vehicle's intention to cross lanes but the current vehicle fails to recognize it, resulting in a collision, thereby improving the safety of autonomous driving. The above threshold values are preferred angle values obtained based on experimental results and do not constitute a limitation of the technical solution disclosed herein.

[0041] As an optional embodiment, generating the lane-crossing intention information of the target vehicle based on the perception information and the lane-crossing intention judgment strategy can include the following methods: determining the lane-crossing scene type corresponding to the target vehicle; determining the target strategy corresponding to the target vehicle based on the lane-crossing scene type, wherein the target strategy is one of the first strategy, the second strategy, and the third strategy; generating the lane-crossing intention information of the target vehicle based on the perception information and the target strategy.

[0042] Optionally, different strategies can be matched to the target vehicle's lane-crossing intention identification process based on the lane-crossing scenario type described by the target vehicle. Since the identification of whether the target vehicle has a lane-crossing intention is gradually relaxed from the first strategy to the second strategy, and from the second strategy to the third strategy, the third strategy can be matched to the lane-crossing scenario type that requires the most attention to driving safety, and the first strategy can be matched to the lane-crossing scenario type that is less concerned with driving safety and more focused on driving smoothness. For example, the target vehicle type can be identified. If the target vehicle is a small car and is driving slowly, the first strategy can be matched to determine whether it has a lane-crossing intention. If the target vehicle is a large car and is driving quickly, the second or third strategy can be matched to determine whether it has a lane-crossing intention.

[0043] As an optional embodiment, the target strategy corresponding to the target vehicle is determined according to the lane crossing scene type, and the following method can be adopted: when the lane crossing scene type is crossing with a turn signal, the first strategy is determined as the target strategy; when the lane crossing scene type is crossing on the line, the second strategy is determined as the target strategy; when the lane crossing scene type is crossing at a bus stop, the third strategy is determined as the target strategy.

[0044] As an optional embodiment, generating lane-crossing intention information of a target vehicle based on perception information and a lane-crossing intention judgment strategy may include the following steps: processing the perception information based on a first strategy, a second strategy, and a third strategy, respectively, and generating lane-crossing intention information for characterizing the presence of a lane-crossing intention of the target vehicle when it is judged that the target vehicle has a lane-crossing intention based on any one of the first strategy, the second strategy, and the third strategy.

[0045] In this optional embodiment, the above first strategy, second strategy and third strategy can be executed in parallel. As long as one of them is met, the target vehicle's intention to cross the lane is directly recalled, and other strategies are exited at the same time to reduce time consumption, ensure the accuracy of judgment and improve judgment efficiency.

[0046] Figure 2 is a schematic diagram of a lane-crossing intention judgment strategy provided according to an optional embodiment of the present disclosure, such as Figure 2As shown, the lane-crossing intention judgment strategy may include a first strategy for judging based on whether the preceding vehicle turns on the turn signal, a second strategy for judging based on whether the preceding vehicle crosses the line, and a third strategy for judging based on whether the preceding vehicle is located at a bus stop scene. The preceding vehicle is the target vehicle mentioned above, and the third strategy based on whether the preceding vehicle is located at a bus stop scene may also include three sub-strategies. When any one of the three sub-strategies is met, it is judged that the target vehicle has a lane-crossing intention and lane-crossing intention information indicating that the target vehicle has a lane-crossing intention is generated to verify the lane-crossing intention prediction result output by the prediction module. The three sub-strategies are the first condition, the second condition, and the third condition provided in the above-mentioned optional embodiment.

[0047] Figure 3 is a schematic diagram of perception information provided according to an optional embodiment of the present disclosure, such as Figure 3 As shown, the perception information can include the speed direction, head direction, S direction speed component and L direction speed component of the target vehicle, where the S direction represents the straight direction and the L direction represents the left direction. The above perception information can be obtained through the perception system of the current vehicle. Figure 3 The right view shows how to calculate the speed angle and the vehicle heading angle. angle_diff represents the angle size. The rectangular boxes in the figure represent lane lines or road lines on the ground.

[0048] According to an embodiment of the present disclosure, a device for predicting a lane-crossing intention is provided for implementing the above-mentioned lane-crossing intention prediction method. Figure 4 is a structural block diagram of a device for predicting lane-crossing intention according to an embodiment of the present disclosure, such as Figure 4 As shown, the lane-crossing intention prediction device 40 includes: a first acquisition module 42, a generation module 44, a second acquisition module 46 and a processing module 48. The lane-crossing intention prediction device 40 is described below.

[0049] A first acquisition module 42 is used to obtain the perception information corresponding to the target vehicle;

[0050] A generating module 44, connected to the first acquiring module 42, is configured to generate lane-crossing intention information of the target vehicle based on the perception information and the lane-crossing intention judgment strategy;

[0051] a second acquisition module 46 connected to the generation module 44 and configured to obtain a prediction result of a lane crossing intention of a target vehicle by a prediction module of a lane crossing prediction model, wherein the prediction module is a machine learning model configured to predict whether the target vehicle has a lane crossing intention by predicting a trajectory of the target vehicle;

[0052] The processing module 48 is connected to the second acquisition module 46 and is used to process the lane crossing intention prediction result according to the lane crossing intention information to obtain a model output result of the lane crossing prediction model, wherein the model output result matches the lane crossing intention information.

[0053] It should be noted here that the above-mentioned first acquisition module 42, generation module 44, second acquisition module 46 and processing module 48 correspond to steps S102 to S108 in the embodiment, and the instances and application scenarios implemented by multiple modules and corresponding steps are the same, but are not limited to the contents disclosed in the above-mentioned embodiment.

[0054] As an optional embodiment, the perception information acquired by the first acquisition module includes at least one of the following: line crossing information, turn signal information, speed angle and vehicle head direction angle; line crossing information is used to characterize whether the target vehicle has crossed the road line, and the road line is the road line close to the current lane included in the lane where the target vehicle is located; turn signal information is used to characterize the turn signal status of the target vehicle; the speed angle is used to characterize the angle between the speed direction of the target vehicle and the road line; the vehicle head direction angle is used to characterize the angle between the vehicle head direction and the road line.

[0055] As an optional embodiment, the lane-crossing intention judgment strategy for generating lane-crossing intention information in the generation module includes at least one of the following: a first strategy, a second strategy, and a third strategy, wherein the first strategy is to judge that the target vehicle has a lane-crossing intention when the target vehicle turns on the turn signal in the direction of the current lane, the speed angle deviates toward the current lane and is greater than a first threshold; the second strategy is to judge that the target vehicle has a lane-crossing intention when the target vehicle passes over the road line, the speed angle deviates toward the current lane and is greater than a second threshold; the third strategy is to judge that the target vehicle has a lane-crossing intention when the target vehicle meets any one of the first condition, the second condition, and the third condition, wherein the first condition includes: any one of the speed angle and the vehicle head direction angle deviates toward the current lane and is greater than the third threshold; the second condition includes: the target vehicle turns on the turn signal in the direction of the current lane and / or passes over the road line, any one of the speed angle and the vehicle head direction angle deviates toward the current lane and is greater than a fourth threshold; the third condition includes: the target vehicle turns on the turn signal in the direction of the current lane and passes over the road line, and both the speed angle and the vehicle head direction angle deviate toward the current lane.

[0056] As an optional embodiment, the generation module includes: a first determination unit, used to determine the lane-crossing scenario type corresponding to the target vehicle; a second determination unit, used to determine the target strategy corresponding to the target vehicle based on the lane-crossing scenario type, wherein the target strategy is one of the first strategy, the second strategy and the third strategy; and a first generation unit, used to generate the lane-crossing intention information of the target vehicle based on the perception information and the target strategy.

[0057] As an optional embodiment, the second determination unit includes: a first determination subunit, used to determine the first strategy as the target strategy when the crossing scene type is crossing with a turn signal; a second determination subunit, used to determine the second strategy as the target strategy when the crossing scene type is crossing on the line; and a third determination subunit, used to determine the third strategy as the target strategy when the crossing scene type is crossing at a bus stop.

[0058] As an optional embodiment, the generation module includes: a second generation unit, which is used to process perception information based on the first strategy, the second strategy and the third strategy respectively, and generate lane-crossing intention information for characterizing the presence of lane-crossing intention of the target vehicle when it is judged that the target vehicle has the intention to cross the lane based on any one of the first strategy, the second strategy and the third strategy.

[0059] As an optional embodiment, the processing module includes: a judgment unit, used to judge whether there is a result matching the lane crossing intention information in the lane crossing intention prediction result; a recall unit, used to recall the prediction result matching the lane crossing intention information from all the prediction results of the prediction module for the target vehicle and use it as the model output result of the lane crossing prediction model when there is no result matching the lane crossing intention information in the lane crossing intention prediction result.

[0060] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0061] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0062] Figure 5 A schematic block diagram of an example electronic device 500 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0063] like Figure 5As shown, the device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. Various programs and data required for the operation of the device 500 can also be stored in the RAM 503. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0064] Various components in device 500 are connected to I / O interface 505, including: an input unit 506, such as a keyboard, mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a magnetic disk, optical disk, etc.; and a communication unit 509, such as a network card, modem, wireless communication transceiver, etc. The communication unit 509 allows device 500 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0065] The computing unit 501 may be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 performs the various methods and processes described above, such as the cross-lane intention prediction method. For example, in some embodiments, the cross-lane intention prediction method may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed onto the device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the cross-lane intention prediction method described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to perform the cross-lane intention prediction method in any other suitable manner (e.g., via firmware).

[0066] Figure 6 is a structural block diagram of an autonomous driving vehicle provided according to an embodiment of the present disclosure, such as Figure 6As shown, the autonomous driving car 60 includes a lane-crossing intention prediction device 40. The lane-crossing intention prediction device 40 can accurately predict whether the target vehicle has a lane-crossing intention and obtain a model output result. The model output result can then be given to the vehicle control system in the autonomous driving car 60, and the vehicle control system completes the control of the autonomous driving car 60 according to the model output result.

[0067] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0068] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0069] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0070] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0071] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0072] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0073] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0074] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for predicting cross-lane intention, comprising: Obtaining the perception information corresponding to the target vehicle; generating lane-crossing intention information of the target vehicle based on the perception information and the lane-crossing intention judgment strategy; Obtaining a prediction result of a lane-crossing intention of the target vehicle from a prediction module of a lane-crossing prediction model, wherein the prediction module is a machine learning model for predicting whether the target vehicle has a lane-crossing intention by predicting a trajectory of the target vehicle; The crossing intention prediction result is processed according to the crossing intention information to obtain the model output result of the crossing prediction model, wherein the model output result matches the crossing intention information. The processing of the crossing intention prediction result according to the crossing intention information to obtain the model output result of the crossing prediction model includes: judging whether there is a result that matches the crossing intention information in the crossing intention prediction result; if there is no result that matches the crossing intention information in the crossing intention prediction result, recalling the prediction result that matches the crossing intention information from all the prediction results of the prediction module for the target vehicle and using it as the model output result of the crossing prediction model.

2. The method according to claim 1, wherein The sensing information includes at least one of the following: line crossing information, turn signal information, speed angle, and vehicle head direction angle; The line crossing information is used to indicate whether the target vehicle has crossed a road line, where the road line is a road line close to the current lane and included in the lane where the target vehicle is located; The turn signal information is used to represent the turn signal status of the target vehicle; The speed angle is used to represent the angle between the speed direction of the target vehicle and the road line; The vehicle head orientation angle is used to represent the angle between the vehicle head orientation of the target vehicle and the road line.

3. The method according to claim 2, wherein: The lane crossing intention judgment strategy includes at least one of the following: a first strategy, a second strategy, and a third strategy, wherein: The first strategy is to determine that the target vehicle has an intention to cross lane when the target vehicle turns on its turn signal in the direction of the current lane and the speed angle deviates toward the current lane and is greater than a first threshold; The second strategy is to determine that the target vehicle has a lane-crossing intention when the target vehicle crosses the road line and the speed angle deviates toward the current lane and is greater than a second threshold; The third strategy is to determine whether the target vehicle has a lane-crossing intention when the target vehicle satisfies any one of the first condition, the second condition, and the third condition, wherein: The first condition includes: either the speed angle or the vehicle head direction angle deviates toward the current lane and is greater than a third threshold; The second condition includes: the target vehicle turns on the turn signal in the direction of the current lane and / or passes the road line, and either the speed angle or the vehicle head direction angle deviates toward the current lane and is greater than a fourth threshold; The third condition includes: the target vehicle turns on the turn signal in the direction of the current lane and crosses the road line, and the speed angle and the vehicle head direction angle are both biased towards the current lane.

4. The method according to claim 3, wherein: The generating, based on the perception information and the lane-crossing intention judgment strategy, the lane-crossing intention information of the target vehicle includes: Determining a lane-crossing scenario type corresponding to the target vehicle; Determining a target strategy corresponding to the target vehicle according to the lane-crossing scenario type, wherein the target strategy is one of the first strategy, the second strategy, and the third strategy; Based on the perception information and the target strategy, lane-crossing intention information of the target vehicle is generated.

5. The method according to claim 4, wherein The determining, based on the lane-crossing scenario type, a target strategy corresponding to the target vehicle includes: When the lane-crossing scenario type is lane-crossing with a turn signal, determining the first strategy as the target strategy; When the lane crossing scenario type is a lane crossing on a line, determining the second strategy as the target strategy; When the lane-crossing scenario type is a lane-crossing at a bus stop, the third strategy is determined as the target strategy.

6. The method according to claim 3, wherein: The generating, based on the perception information and the lane-crossing intention judgment strategy, the lane-crossing intention information of the target vehicle includes: The perception information is processed based on the first strategy, the second strategy and the third strategy respectively, and when it is determined that the target vehicle has an intention to cross the lane based on any one of the first strategy, the second strategy and the third strategy, lane-crossing intention information is generated to characterize that the target vehicle has an intention to cross the lane.

7. A lane-crossing intention prediction device, comprising: A first acquisition module is used to obtain the perception information corresponding to the target vehicle; A generating module, configured to generate lane-crossing intention information of the target vehicle based on the perception information and the lane-crossing intention judgment strategy; A second acquisition module is used to obtain a prediction result of the lane crossing intention of the target vehicle by a prediction module of the lane crossing prediction model, wherein the prediction module is a machine learning model used to predict whether the target vehicle has the lane crossing intention by predicting the trajectory of the target vehicle; A processing module is used to process the lane-crossing intention prediction result according to the lane-crossing intention information to obtain a model output result of the lane-crossing prediction model, wherein the model output result matches the lane-crossing intention information. The processing module includes: a judgment unit, used to judge whether there is a result that matches the lane-crossing intention information in the lane-crossing intention prediction result; and a recall unit, used to recall the prediction result that matches the lane-crossing intention information from all prediction results of the prediction module for the target vehicle when there is no result that matches the lane-crossing intention information in the lane-crossing intention prediction result and use it as the model output result of the lane-crossing prediction model.

8. The device according to claim 7, wherein The perception information acquired by the first acquisition module includes at least one of the following: line crossing information, turn signal information, speed angle, and vehicle head direction angle; The line crossing information is used to indicate whether the target vehicle has crossed a road line, where the road line is a road line close to the current lane and included in the lane where the target vehicle is located; The turn signal information is used to represent the turn signal status of the target vehicle; The speed angle is used to represent the angle between the speed direction of the target vehicle and the road line; The vehicle head orientation angle is used to represent the angle between the vehicle head orientation of the target vehicle and the road line.

9. The device according to claim 8, wherein The lane-crossing intention judgment strategy used in the generation module to generate the lane-crossing intention information includes at least one of the following: a first strategy, a second strategy, and a third strategy, wherein: The first strategy is to determine that the target vehicle has an intention to cross lane when the target vehicle turns on its turn signal in the direction of the current lane and the speed angle deviates toward the current lane and is greater than a first threshold; The second strategy is to determine that the target vehicle has a lane-crossing intention when the target vehicle crosses the road line and the speed angle deviates toward the current lane and is greater than a second threshold; The third strategy is to determine whether the target vehicle has a lane-crossing intention when the target vehicle satisfies any one of the first condition, the second condition, and the third condition, wherein: The first condition includes: either the speed angle or the vehicle head direction angle deviates toward the current lane and is greater than a third threshold; The second condition includes: the target vehicle turns on the turn signal in the direction of the current lane and / or passes the road line, and either the speed angle or the vehicle head direction angle deviates toward the current lane and is greater than a fourth threshold; The third condition includes: the target vehicle turns on the turn signal in the direction of the current lane and crosses the road line, and the speed angle and the vehicle head direction angle are both biased towards the current lane.

10. The device according to claim 9, wherein The generation module includes: A first determining unit is configured to determine a lane-crossing scenario type corresponding to the target vehicle; a second determining unit, configured to determine a target strategy corresponding to the target vehicle according to the lane-crossing scenario type, wherein the target strategy is one of the first strategy, the second strategy, and the third strategy; The first generating unit is configured to generate lane-crossing intention information of the target vehicle based on the perception information and the target strategy.

11. The device according to claim 10, wherein The second determining unit includes: a first determining subunit, configured to determine, when the lane-crossing scenario type is lane-crossing with a turn signal, that the first strategy is the target strategy; a second determining subunit, configured to determine, when the lane crossing scenario type is a lane crossing on a line, that the second strategy is the target strategy; The third determining subunit is configured to determine the third strategy as the target strategy when the lane crossing scenario type is a lane crossing at a bus stop.

12. The device according to claim 11, wherein The generation module includes: The second generation unit is used to process the perception information based on the first strategy, the second strategy and the third strategy respectively, and generate lane-crossing intention information for characterizing that the target vehicle has the intention to cross the lane when it is judged that the target vehicle has the intention to cross the lane based on any one of the first strategy, the second strategy and the third strategy.

13. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the cross-lane intention prediction method according to any one of claims 1 to 6.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to enable the computer to execute the lane-crossing intention prediction method according to any one of claims 1 to 6. 15 . A computer program product, comprising a computer program, wherein when executed by a processor, the computer program implements the method for predicting a lane-crossing intention according to claim 1 .

16. An autonomous vehicle, wherein: The autonomous driving vehicle includes the lane-crossing intention prediction device according to any one of claims 7 to 12.

Images